The voice of WALL-E

Today’s article is about Disney-Pixar’s animation film Wall-E and the memorable voice of the main little robot character. According to sound designer Ben Burtt, the robot voices are “like a toddler […] universal language of intonation. ‘Oh’, ‘Hm?’, ‘Huh!’, you know?”[44].

Ben Burtt explains how he did the voice of Wall-E here: “You start with the human voice input and record words or sounds and then it is taken into a computer and I worked out a unique program which allowed me to deconstruct the sound into its component parts.

I could reassemble the Wall-E vocals and perform it with a light pen on a tablet. You could change pitch by moving the pen or the pressure of the pen would sustain or stretch syllables or consonants and you could get an additional level of performance that way, kind of like playing a musical instrument.


Voices are the hardest because the audience listens to them with much more critical ears than sound effects. We are all experts at interpreting the nuances of speech, so anything that might be interpreted as a vocal or expression the audience really listens carefully.”

Here’s a little example of what I’ve been able to recreate in Kyma with my own voice:


FFT Synthesis/Resynthesis

  • Additive synthesis parameters in a discrete-time implementation can be determined using the Fast Fourier Transform (FFT).
  • The analyzed time-domain signal is split into blocks orΒ  ‘frames”, each of which is processed using the FFT (referred to as the Short-Time Fourier Transform (STFT).
  • The STFT provides a means for joint time-frequency analysis.
  • As well, a time-domain signal can be resynthesized using the Inverse Fast Fourier Transform (IFFT). The resulting IFFT frames are “assembled” using overlap-add techniques.
  • With improvements in computer processing speed, it is now possible to perform IFFT resynthesis in real time.
  • FFT/IFFT synthesis lends itself well to sound transformations, such as time-stretching and pitch scaling.

Here’s a very short and simple explanation of the Kyma patch that you can see at the bottom of this page:

As Ben Burtt simplified:

We all know how pictures are pixels now and you can rearrange pixels to change the picture. You kind of do the same thing with sound.


  • FFT Synthesis/Resynthesis patch in Kyma:

    • Disjt
    • August 27th, 2010

    Would it be possible to share your Kyma patch?
    Of course I would understand if not πŸ˜‰

    • admin
    • August 27th, 2010

    I believe anybody can find his own way to do this, look at the patch above.

    “The wonderful thing about working with Kyma is that, like any good art form, the more you do it, the more ideas it generates. β€” Hamilton Sterling”

  1. Good info

  2. For the help please use

    • rumplestiltskin
    • September 15th, 2010

    Very nice. How did you get the time index – repeating loop as an input to the Tau Player? Is that a constant?

  3. No, it’s a repeatingFullRamp function. But you might get some good results with a hot value. πŸ˜‰

    • Ruppa
    • July 9th, 2012

    Patch would be very handy! You just releases super awesome whoosh patch!

  4. Jean, I cannot figure this one out. I have only been using Kyma a couple of months, so I am still very new. What is ‘Log Magnitude,’ and what does it do? I am unclear on some of these sounds, as I have no idea what to set them to. Any help?

  5. Hey Kyle, I use the Log Magnitude to scale and offset the extracted frequency content while keeping the amplitude on the other branch of the patch as is. It’s just a way to separate them both and process them independently, then combine them later using the SpectrumLogToLinear. If you only had Kyma for two months, it’s normal that you feel that a lot of things are confusing and sometimes difficult to understand. Use Kyma Revealed as much as you can, I read the book twice and I keep using it all the time as a reference. Have fun!

    • Martin
    • January 27th, 2013

    Hi Jean-Edouard,
    I am also wondering as Kyle do, where did you get this “log magnitude” sound? I searched for it it and never find it in the prototype.
    Thanks for your help!
    (et bravo pour ton super travail!)

    • Martin
    • January 27th, 2013

    Sorry for this question, I ask Kyma support and it seems logmagnitude has been remove from the prototypes system in the last version…


  6. Hi,

    I discovered this post when I was checking if a google search would come up with my recently released tutorial which is about the same Sound – check it out on my blog:

    See you!

  7. Oh, and by the way, Kyle, the LogMagnitude is needed because the output of the Live Spectral Analysis is linear. It makes sense to process the frequencies in a logarithmic way because thats the way we perceive them. As a workaround I think you could use some Math prototype to do the conversion.. Or use the spectruminRAM with a spectrum file which outputs the frequencies logarithmic.


  1. No trackbacks yet.